This repository contains a set of fat, developer-friendly, general purpose NodeJS images for Docker.
- 2 variants available:
standalone
andapache
(for serving an SPA without NodeJS server as backend) - For each variant, you can select the
build
sub-variant that contains essential build tools (make, gcc, etc.) that can be needed to compile native extensions - Images are bundled with cron. Cron jobs can be configured using environment variables
- Everything is done to limit file permission issues that often arise when using Docker. The image is actively tested on Linux, Windows and MacOS
- Base image is Debian Bullseye (with more variants to come)
Note: we do not tag minor releases of NodeJS, only major versions. You will find for example an image for NodeJS 18.x, but no tagged image for NodeJS 18.2. This is because NodeJS follows SemVer and we believe you have no valid reason to ask explicitly for 18.2. When 18.3 is out, you certainly want to upgrade automatically to this minor release since it is backward compatible.
Images are automatically updated when a new patch version of NodeJS is released, so the NodeJS 18.x image will always contain the most up-to-date version of the NodeJS 18.x branch.
Example with standalone:
$ docker run -it --rm --name my-running-script -v "$PWD":/usr/src/app thecodingmachine/nodejs:v2-18-bullseye node your-script.js
Example with Apache:
$ docker run -p 80:80 --name my-apache-app -v "$PWD":/var/www/html thecodingmachine/nodejs:v2-18-apache-bullseye
Example with Apache + Node 18.x in a Dockerfile:
Dockerfile
FROM thecodingmachine/nodejs:v2-18-apache-bullseye
COPY src/ /var/www/html/
RUN yarn install
RUN yarn run buld
The working directory (the directory in which you should mount/copy your application) depends on the image variant you are using:
Variant | Working directory |
---|---|
standalone | /usr/src/app |
apache | /var/www/html |
For the apache variant, you can change the document root of Apache (i.e. your "public" directory) by using the
APACHE_DOCUMENT_ROOT
variable:
# The root of your website is in the "public" directory:
APACHE_DOCUMENT_ROOT=public/
You can enable/disable Apache extensions using the APACHE_EXTENSION_[extension_name]
environment variable.
For instance:
version: '3'
services:
my_app:
image: thecodingmachine/nodejs:v2-18-apache-bullseye
environment:
# Enable the DAV extension for Apache
APACHE_EXTENSION_DAV: 1
# Enable the SSL extension for Apache
APACHE_EXTENSION_SSL: 1
As an alternative, you can use the APACHE_EXTENSIONS
global variable:
APACHE_EXTENSIONS="dav ssl"
Apache modules enabled by default: access_compat, alias, auth_basic, authn_core, authn_file, authz_core, authz_host, authz_user, autoindex, deflate, dir, env, expires, filter, mime, mpm_prefork, negotiation, reqtimeout, rewrite, setenvif, status
Apache modules available: access_compat, actions, alias, allowmethods, asis, auth_basic, auth_digest, auth_form, authn_anon, authn_core, authn_dbd, authn_dbm, authn_file, authn_socache, authnz_fcgi, authnz_ldap, authz_core, authz_dbd, authz_dbm, authz_groupfile, authz_host, authz_owner, authz_user, autoindex, buffer, cache, cache_disk, cache_socache, cgi, cgid, charset_lite, data, dav, dav_fs, dav_lock, dbd, deflate, dialup, dir, dump_io, echo, env, ext_filter, file_cache, filter, headers, heartbeat, heartmonitor, ident, include, info, lbmethod_bybusyness, lbmethod_byrequests, lbmethod_bytraffic, lbmethod_heartbeat, ldap, log_debug, log_forensic, lua, macro, mime, mime_magic, mpm_event, mpm_prefork, mpm_worker, negotiation, proxy, proxy_ajp, proxy_balancer, proxy_connect, proxy_express, proxy_fcgi, proxy_fdpass, proxy_ftp, proxy_html, proxy_http, proxy_scgi, proxy_wstunnel, ratelimit, reflector, remoteip, reqtimeout, request, rewrite, sed, session, session_cookie, session_crypto, session_dbd, setenvif, slotmem_plain, slotmem_shm, socache_dbm, socache_memcache, socache_shmcb, speling, ssl, status, substitute, suexec, unique_id, userdir, usertrack, vhost_alias, xml2enc
Ever faced file permission issues with Docker? Good news, this is a thing of the past!
If you are used to running Docker containers with the base NodeJS image, you probably noticed that when running commands
(like yarn install
) within the container, files are associated to the root
user. This is because the base user
of the image is "root".
When you mount your project directory into /var/www/html
or /usr/src/app
, it would be great if the default user used by Docker could
be your current host user.
The problem with Docker is that the container and the host do not share the same list of users. For instance, you might
be logged in on your host computer as superdev
(ID: 1000), and the container has no user whose ID is 1000.
The thecodingmachine/nodejs images solve this issue with a bit of black magic:
The image contains a user named docker
. On container startup, the startup script will look at the owner of the
working directory (/var/www/html
for Apache, or /usr/src/app
for standalone). The script will then assume that
you want to run commands as this user. So it will dynamically change the ID of the docker user to match the ID of
the current working directory user.
Furthermore, the image is changing the Apache default user/group to be docker/docker
(instead if www-data/www-data
).
So Apache will run with the same rights as the user on your host.
The direct result is that, in development:
- Your NodeJS application can edit any file
- Your container can edit any file
- You can still edit any file created by Apache or by the container in CLI
You can set up CRON jobs using environment variables too.
To do this, you need to configure 3 variables:
# configure the user that will run cron (defaults to root)
CRON_USER=root
# configure the schedule for the cron job (here: run every minute)
CRON_SCHEDULE=* * * * * * *
# last but not least, configure the command
CRON_COMMAND=yarn run stuff
By default, CRON output will be redirected to Docker output.
If you have more than one job to run, you can suffix your environment variable with the same string. For instance:
CRON_USER_1=root
CRON_SCHEDULE_1=* * * * * * *
CRON_COMMAND_1=yarn run stuff
CRON_USER_2=www-data
CRON_SCHEDULE_2=* * * * * * *
CRON_COMMAND_2=yarn run other-stuff
Important: The cron runner we use is Supercronic and not the orginial "cron" that has a number of issues with containers. Even with Supercronic, the architecture of cron was never designed with Docker in mind (Cron is way older than Docker). It will run correctly on your container. If at some point you want to scale and add more containers, it will run on all your containers. At that point, if you only want to run a Cron task once for your application (and not once per container), you might want to have a look at alternative solutions like Tasker or one of the many other alternatives.
You can launch commands on container startup using the STARTUP_COMMAND_XXX
environment variables.
This can be very helpful to install dependencies or apply database patches for instance:
STARTUP_COMMAND_1=yarn install
STARTUP_COMMAND_2=yarn run watch &
As an alternative, the images will look into the container for an executable file named /etc/container/startup.sh
.
If such a file is mounted in the image, it will be executed on container startup.
docker run -it --rm --name my-running-script -v "$PWD":/usr/src/myapp -w /usr/src/myapp \
-v $PWD/my-startup-script.sh:/etc/container/startup.sh thecodingmachine/nodejs:v2-18-bullseye node your-script.js
If your NodeJS project as a dependency on a package stored in a private GIT repository,
your yarn install
commands will not work unless you register your private key in the container.
You have several options to do this.
This option is the easiest way to go if you are using the image on a development environment.
docker-compose.yml
version: '3'
services:
my_app:
image: thecodingmachine/nodejs:v2-18-bullseye
volumes:
- ~/.ssh:/home/docker/.ssh
Look at this option if you are building a Dockerfile from this image.
The first thing to do is to get the signature of the server you want to connect to.
$ ssh-keyscan myserver.com
Copy the output and put it in an environment variable. We assume the content is stored in $SSH_KNOWN_HOSTS
.
Now, let's write a Dockerfile.
Dockerfile
FROM thecodingmachine/nodejs:node1°
ARG SSH_PRIVATE_KEY
ARG SSH_KNOWN_HOSTS
# Let's register the private key
RUN ssh-add <(echo "$SSH_PRIVATE_KEY")
# Let's add the server to the list of known hosts.
RUN echo "$SSH_KNOWN_HOSTS" >> ~/.ssh/known_hosts
Finally, when triggering the build, you must pass the 2 variables as build arguments:
$ docker build -t my_image --build-arg SSH_PRIVATE_KEY="$SSH_PRIVATE_KEY" --build-arg SSH_KNOWN_HOSTS="$SSH_KNOWN_HOSTS" .
If you plan to use this image in Kubernetes, please be aware that the image internally uses sudo
. This is because the
default user (docker
) needs to be able to edit NodeJS config files as root
.
Kubernetes has a security setting (allowPrivilegeEscalation
) that can disallow the use of sudo
. The use of this flag
breaks the image and in the logs, you will find the message:
sudo: effective uid is not 0, is /usr/bin/sudo on a file system with the 'nosuid' option set or an NFS file system without root privileges?
Please be sure that this option is never set to false:
apiVersion: v1
kind: Pod
# ...
spec:
containers:
- name: foobar
image: thecodingmachine/nodejs:v2-18-bullseye
securityContext:
allowPrivilegeEscalation: true # never use "false" here.
The Dockerfiles are generated from a template using Orbit.
If you want to modify a Dockerfile you should instead edit the utils/Dockerfile.blueprint
and then run the command:
$ orbit run generate
This command will generate all the files from the "blueprint" templates.
You can then test your changes using the build-and-test.sh
command:
TAG=v2-18-bullseye VARIANT=18-bullseye ./build-and-test.sh